[1230] in Kerberos_V5_Development
release testing status
daemon@ATHENA.MIT.EDU (Sam Hartman)
Thu May 23 23:51:21 1996
Date: Thu, 23 May 1996 23:51:10 -0400
From: Sam Hartman <hartmans@MIT.EDU>
To: krbdev@MIT.EDU
First, let me start off by discussing issues I haven't tested
much that I believe are likely to cause problems. Most of these are
obvious and well-known, but if you have time before Beta6, feel free
to spend some time working on checking these:
* Modification to login.c have not been tested much at all by MIT;
I'm not sure how much Cygnus did before contributing the code. Some
of the code looks rough, and Ken agrees that login could benefit
from someone's attension.
* Following the directions to set up a realm and see what
breaks/works.
* Following direction to set up kpropd and making sure slave servers
works.
* Using the krb5 kadmind. This works well enough for make check, but
that is saying very little.
* Interoperability testing with older version of krb5.
This week I have been pounding on the krb5 locker, confirming
that I could build everything and that it passed make check. So far,
the following platforms appear to work:
* Solaris
* NETBSD
* AIX
* Alpha
* SunOS ((not ftpd--se below)
I have not confirmed Ultrix do to lack of time on a decstation
since my last change to get NetBSD working. It was working before, so
it should probably still work.
Known problems:
* krb524init fails on Alpha, probably indicating some problem getting
local addresses for tickets. Basically, if a client cannot get local
addresses, it requests tickets with null addresses. These work fine
for krb5 but cannot be converted to krb4 by lib krb524. I will try to
look into this, but no high expectation of success.
* AIX4 fails to build shared libraries because of the way top-level
configure.in works. I know about two ways of fixing this but haven't
implemented them do to lack of time. Note that AIX4 is not one of the
platforms I normally test but I just happen to know about this.
* AIX4 crashes causing a kernel panic or system hang in klogind and
telnetd. I believe this is a kernel bug with the revoke system call
called on ptys but am not sure. This should be documented for Beta6;
I'll try to do so.
* SUNOS failed the ftp test because it claimed I wasn't authorized to
log in as hartmans, even though the appropriate principal
(hartmans@KRBTEST.COM) was in my .k5login. Note that something is
strange because the rsh tests (which requires the same things out of
.k5login) worked fine. I haven't looked in to this much, and it may
just be an Athena installation problem.
* NETBSD AFS deals poorly with locking, so run make check on a FFS
filesystem.
